Towards automatic household food waste measurement with the smart compost bin

Abstract

Rising amounts of global food waste constitute an environmental, social, and economic disaster. Over 60% of food waste is generated by household consumers, yet we currently lack reliable means to quantify this waste. This lack of a robust household food waste measurement methodology has hindered global efforts towards meeting United Nations Sustainable Development Goal 12.3, which calls on nations to reduce global food waste by 50% by 2030. The paucity of publicly available commingled food waste image datasets has prohibited researchers from leveraging advances in computer vision for the task of automatic food waste measurement. We present an AI-enabled compost bin that measures kitchen compost waste by collecting waste images alongside environmental and weight data. We streamline the food waste annotation routine, combining weighing and imaging into a single step while enabling hands-free verbal waste labeling at time of measurement via automatic speech recognition. We provide a companion smartphone application which both provides waste analytics to users and serves as a platform for collecting instance-level food waste image annotations. To motivate a forthcoming field study—in which we will deploy 50 smart compost bins to curate a large, novel dataset of commingled food waste images—we present a computer vision model for food image recognition and demonstrate the inability of such an approach to transfer to the food waste domain.

The authors' abstract, as published at the source. Discover Sustainability, 2026 · DOI ↗